[gmx-users] Re: What is epsilon_r?

[Andrew DeYoung]

Dr. van der Spoel,

Thank you for your time!  So it seems that epsilon_r is just a "fudge
factor" and has nothing to do with the dielectric constant at the boundary
of the "PME/Ewald sphere."  Is that dielectric constant actually
epsilon_surface?  I am not very familiar at all with other MD codes, but a
colleague recently told me that some other codes use a default dielectric
constant of infinity at the surface.  However, in Gromacs, the default value
of epsilon_surface is 0, which corresponds (I guess) to "no dielectric" at
the surface.  Is this true?

Also, to complicate things further, I am using a pseudo 2-D PME summation
(ewald_geometry = 3dc).  I have read in the manual entry for epsilon_surface
that epsilon_surface "does not affect the slab 3DC variant of the long range
corrections."  If you have time, what does this mean?  Does this mean that,
if I use ewald_geometry = 3dc, then there is "no dielectric" at the surface
(the medium surrounding the "Ewald sphere")?

Thank you!

Andrew DeYoung
Carnegie Mellon University

> epsilon_r should never be used with a regular force field because the
> balance between coulomb and other terms will be wrong. maybe we should
> remove the term.